Add Foundry showmap relscore pipeline

[figtracer]

PR #11 is fragile because SCFuzzBench artifacts are uploaded per runner/fuzzer, while Foundry showmap output can be nested by approach, suite, test, and trial. This adds the missing normalization layer before scoring and runs normalized campaigns through the differential-coverage package for both relscore and relcov:

• raw uploaded showmap/ trees are preserved for debugging
• Foundry replay keeps normal FOUNDRY_TEST_ARGS such as RPC/thread flags, while script-owned showmap flags win during replay
• showmap replay lets forge resolve corpus directories from Foundry config unless FOUNDRY_SHOWMAP_CORPUS_DIR/SCFUZZBENCH_CORPUS_DIR is explicitly set
• analysis materializes canonical approach/trial.txt campaign directories under showmap_campaigns/
• showmap_campaigns/combined/<approach>/<trial>.txt is the stable overall differential-coverage input
• showmap_campaigns/by_test/<suite-test>/<approach>/<trial>.txt keeps per-test drill-down campaigns
• differential_coverage_relscores.csv records relscore values from differential-coverage
• differential_coverage_relcov.csv records pairwise relcov values from differential-coverage
• showmap_campaign_manifest.json records included/skipped artifacts so missing or invalid showmap files are visible
• filtered analysis applies the same fuzzer exclusions to showmap scoring as to event/throughput/progress CSVs

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	pypi/​differential-coverage@​1.1.0

View full report

[mablr]

Please check if we need to update benchmark-release.yml, to copy showmap_campaign_manifest.json and showmap_campaigns/ produced by analysis/analyze.py.

[mablr]

sgtm